My ideal for a reformed electoral system

I’ve written about electoral reform in Canada a couple times. It’s certainly a topic that interests me. Thing is, the only thing I’ve done in my previous posts on the topic is point out the problems. I never really detailed a particular solution. I’d like to change that.

Pointing out problems is a good start, but too often people will go no further. If you’re not going to offer solutions, then just pointing out problems repeatedly is not only a waste of time, it’s an annoying distraction. We know the world isn’t perfect, thank you. Most of us are quite well aware of its problems. Those that we can’t fix, we try to tolerate – having someone rub them in our faces doesn’t make that easier.

I take that to heart, so whenever I speak out against some problem in Canadian society, I make sure I have a solution to offer ready at hand. Electoral reform is no exception.

Obviously the solution I have in mind, once we scrap our current first-past-the-post electoral system, is a system of proportional representation. I have explained before that the only way to have a fair and democratic electoral system is if that system is a proportional one. But it’s not good enough to have a only a vague notion of what the solution is. I like to have a fairly concrete solution in mind.

So what I’m going to do in this post is describe, in detail, the electoral system I would like to see implemented as a replacement for our current FPTP system. As I go along, I will explain why I have chosen each feature, and I’ll give some realistic examples of how voters might use the system to express their will. I don’t intend to specify precisely, down to the smallest detail. But I’d like to specify enough that it should be clear that the details are just that – details – which can be handled.

But before I begin, I want to make it clear: This is just my personal ideal for a replacement electoral system. It is not the system advocated by proponents of proportional representation in general. And it is not the only replacement system that I would support. In fact, an argument could be made that my system might be a little too complicated (though I disagree). You can think of what I’m doing here as merely putting a proposal on the table, with the expectation that it should be discussed, and possibly improved with suggestions offered by others. (In fact, I’d love to have suggestions in the comments section.)

With that cleared up, I present to you: Indi’s ideal electoral system.

The basic idea: a mixed-member proportional system

Proportional representation has been recommended at just about every level of government, right across Canada. And with only a handful of exceptions, the proportional system proposed is a mixed-member proportional system. I agree with that consensus; I think MMP is the best system, at least for Canada at the federal level, and for Ontario as well. (I mention Ontario specifically because it’s my province, which I know well. I’m not going to speak for other provinces or territories.)

There are three main options for PR:

  • Mixed-member (MMP)
  • Party list
  • Single transferable vote (STV)

The idea behind single transferable vote is that instead of ridings electing a single representative, each riding will have multiple winners. For example, you could have a riding with 15 candidates and 4 winners. STV has been used for almost a hundred years in Ireland, and is used sporadically around the Commonwealth (Malta, a bunch of Australian states and territories use it). It was proposed for use in British Columbia in 2005 and 2009, and came very close to passing in 2005.

I’m not going to bother to explain the mechanics of STV – if you’re interested, you can research it yourself. I’m just going to explain why I don’t find it appealing.

First, the idea of huge, multi-member ridings seems problematic. Canada is vast, and very sparsely populated. We already have the biggest electoral districts in the world… can you imagine making them bigger? You can’t really expect a representative to properly represent a district that big. Worse, Canada’s population density is very patchy – you could end up with some areas that have ridings only slightly larger than the current ridings (like around Toronto), and other areas that have ridings bigger than New Brunswick (like around Cambridge Bay, Nunavut).

Second, I don’t believe the idea maps logically to our system of government. I’ll explain what I mean by that in a bit.

And third, it’s too easy for the government in power to fiddle with things. This has actually been a problem in Ireland. If you want to minimize the proportionality threshold, all you have to do is lower the number of winners in a riding.

STV does have some interesting advantages – like that it essentially eliminates the need for parties; independents have as good as shot as anyone else. But I don’t think they apply well to the Canadian political system.

The next option is the party list system, which comes in many flavours. The key concept is that you vote for parties… not people. In one version – the “closed list” version – each party prepares a list of candidates, in order. Voters vote for parties, and then each party is allocated a number of seats determined by the proportion of votes they got. If a party won N seats, the first N people on their list get elected (so parties will logically put their leaders up at the top). Other flavours of party list voting – “open list” flavours – allow voters to pick specific people on a party list. That way, they’re not stuck with the top N candidates the party chose – when the party wins N seats, they will be given to the top N candidates from that party that the voters chose. There is also a version that uses districts to make party lists, but it’s a bit odd because it’s possible that a district will have no winners, while another district may have multiple winners.

The closed list flavour is used in Italy and Israel (that alone should be enough to turn you off the idea – those two countries are the ones always mentioned when people want to argue against PR). Open list flavours are used in most Scandinavian countries, Brazil, and various countries throughout Europe. The district list version isn’t widely used.

The party list system isn’t bad, in the context of a system that would work for Canada. But while it would “work” on one level – that being the structure and functioning of Parliament itself – it kinda misses the point on another. Our parliament is not merely a bunch of seats that must be filled with asses in order to get a voting population for bills. Those asses in those seats are not merely supporters of their various parties, they are also supposed to be representatives for their ridings.

This is what I was getting at earlier when I was talking about how STV doesn’t map “logically” to our system of government. When we go to the polls and vote, we are really making two choices:

  1. We are choosing which party we want to form the government; and
  2. We are choosing who we want to represent us (that is, our district) to that government.

Our current electoral system conflates those two things into a single vote – which is just one of its many problems. Party list voting basically takes away the second option completely. STV does give multiple votes, but they don’t map logically to those two things we want to accomplish when we’re voting.

Which brings us to the third option for a proportional system: mixed-member proportional representation.

MMP is sort of a hybrid between party list proportional representation and our current system (non-proportional, single-winner-per-district). Or another way of looking at is to start with our current system, and then extend that with a party list to get proportionality.

MMP maps perfectly to what we are really doing when we vote in Canada. Every voter gets two votes (as you can see in the German ballot image below, which is a ballot from an election that uses an MMP system):

  1. One vote is for the party you want to lead the government.
  2. The other vote is for a member of that government who will represent your district’s concerns to it.
[Image of a ballot from a German federal election. It is divided into two sides. One side lists the names and parties of eight candidates for the seat of the district that the ballot is from. The other side lists all fourteen political parties that contested the election.]

Ballot from an election for the German Bundestag, which uses a mixed-member proportional system. You can see the candidate list on the left, and the party list on the right.

The way it works is also quite simple to understand. First, the winner the vote in every district is elected – just like how it works today. But what happens next is that there are a bunch of extra seats that still need to be filled. They are pulled from party lists to add to the members already elected, until the proportions seated match the proportions desired in the party list vote.

This makes voting really intuitive, and more flexible for Canadians. Suppose I want the Green Party to form the government… but at the same time, the Green Party representative in my riding is a total tool – the person I want to represent me to government is actually the Conservative candidate. Under the current system, I’m screwed one way or the other – either I vote Green and accept I’ll have a shitty local representative, or I vote Conservative and get a great representative at the cost of possibly helping create a Conservative government. With the MMP system, the problem goes away: I give the representative vote to the Conservative candidate, but select the Green Party in the party list. I can express my desires exactly.

And of course, if the Green candidate in my district is good, I can vote for them and vote for Green in the party list.

This also makes independents viable options. If I like the independent candidate in my riding, I can vote for them and still have the power to choose which party I want to govern with my second vote.

Changing to a MMP system could be nearly painless. We could, in theory, leave all the current districts as they are, and just add a bunch of new seats to be filled out by the party list votes to achieve proportionality. So we’d start with the existing 308 seats (or the 338 we’ll have in the 2015 election), then add, say, 100 new list seats. Then the only thing we’d need to change is the ballots for the next election (and, of course, we’d need tot educate the voters). That’s all it would take.

However, a more practical option might be to redistribute the districts a bit to make fewer, but larger, districts. The main reason why this isn’t a bad idea is that under a MMP system, it’s a lot harder for parties to just parachute candidates into districts and have them get elected. If the locals think the candidate doesn’t really represent them, they won’t vote for them… but unlike under the current system where that would mean “throwing away” the vote they would have wanted to give to your party, they can still give your party the list vote. It’s no loss to them, and it’s no loss to the party either – they’ll get the same proportion of seats regardless of whether one of those seats is in that district or not. So if you, as a candidate, want to win a seat in a riding… you damn well best not phone it in and hope to coast on the party’s success. You’d better actually be a good representative for that riding.

So MMP would make it much more important for candidates to really represent the district – they can’t just phone it in and trust that the party name will be enough to get them elected. That makes it reasonable to make slightly bigger districts. So rather than starting from 308 district seats then adding some amount of list seats, we could start from say 200 district seats, then add 200 list seats – for a total of 400 seats… slightly larger than the current seat count, but not by much. (I’m not going to make a concrete suggestion for the number of seats. It would depend on population size, distribution, and how big a Parliament we want. It would require a bit of study that I am not prepared to do for this thought experiment.)

Although it means making two choices at voting time – rather than one, like we do now – and although it probably means more MPs, I think the MMP model maps most closely to what voters are trying to accomplish by voting in Canada. (At least at the federal level, and the provincial level in Ontario.)

So that’s the basic, overall system I’d suggest: A mixed-member proportional system, probably with slightly fewer and slightly bigger districts than we have now to accommodate the additional proportional list seats. I don’t really see a need for open lists, so I’d suggest a closed list system, but on this point I could be swayed easily.

District-level voting: Alternative Vote

The MMP system is essentially made up of two “subsystems”. One to determine the district seat winners, then another to determine the list seat winners and get proportionality.

The first part is exactly like our current system. And in fact, most places that use MMP use FPTP in the district seat elections. The reasoning is that it’s simple, and that all the disproportionality problems go away with the second stage list seat allotment, so who cares, right?

Well, I care.

Here’s my position: We shouldn’t half-ass our electoral system. FPTP is completely broken for multi-winner situations – like our Parliament… but it’s also broken for single winner situations. Whenever there are more than two candidates in a single-winner election, FPTP fails to produce the sensible result.

What does produce the sensible result for single-winner elections is alternate vote (also known as instant-runoff voting).

Let me make clarify that I do not support AV for use as the overall solutionsomething that the Liberal party has hinted at (they call it “preferential ballot”). While AV produces good results for a single-winner situation – like the contest in a single riding – it does not work for multi-winner situations – like our Parliament.

However, with the two-stage design of a mixed-member system, it could be used effectively as the first stage, and the second stage will handle the proportionality problems.

As far as I’m aware, everywhere that uses MMP uses FPTP as the first stage. Why am I going in a different direction? Let me give you an example.

Suppose you had a riding where 66% of voters despised Candidate A, but were evenly split between Candidate B and Candidate C. With FPTP, you’d get results like this:

  • 34% Candidate A (wins)
  • 33.5% Candidate B
  • 32.5% Candidate C

Despite the fact that almost two thirds of voters voted against Candidate A, they win. FPTP is terrible.

In a AV system, the voters who hate Candidate A but support either Candidate B or Candidate C can give a second vote for the other of those two candidates – to make it clear that they don’t want Candidate A. Most voters for Candidate B would give it rank 1, then give rank 2 to Candidate C. Similarly, voters for Candidate C would likely give their second choice to Candidate B. Let’s assume that all voters for Candidate B and C gave their second choice to the other candidate.

In the first round count, you’d get:

  • 34% Candidate A
  • 33.5% Candidate B
  • 32.5% Candidate C

Candidate C has the fewest votes, so ze’s eliminated, and the second-choice votes on each of its ballots are counted. All of Candidate C voters’ second choice is for Candidate B, so:

  • 34% Candidate A
  • 66% Candidate B (wins)

Candidate B now has 50%+1, so they win. And their win is the actual will of the people. 66% of voters didn’t want Candidate A – they wanted Candidate B or Candidate C. Some of them would have preferred Candidate C, but their second choice was Candidate B, so they’re content. The results reflect the reality of the people’s desire.

So by using AV in each of the elections for the district seat representatives, the winner in each riding is the actual person the majority of the riding wants as representative. That’s what we want.

I’ll do an example later in the post to show how it works. It’s not complicated at all, though it’s obviously more complicated than the FPTP method.

Of course, that’s not the entire solution. AV will select the right candidate district-by-district, but can still leave us with a disproportionate mess overall in Parliament. That’s where the other half of the MMP system comes in.

Party list selection: the Sainte–Laguë method

So the first part of the election process is the selection of the district representatives – more or less the situation we have now (though already improved significantly by using AV instead of FPTP). This isn’t quite the whole story, because it may be the case that while most districts wants representatives from Party A as their representatives, they don’t want them to form the government. Under our current system (or if we just replaced FPTP with AV), we’d be shit-outta-luck. But MMP allows us to fix that problem.

I’m going to use some actual Canadian federal election data here, but fudge it a little bit. We don’t actually have a MMP system, so there was no second, party list vote in 2011. What I’m going to do to make up for that is pretend the popular vote is the party list vote.

So in 2011, there were 308 district seats, allocated as follows:

  • Conservative: 166 (53.9%)
  • NDP: 103 (33.4%)
  • Liberal: 34 (11.0%)
  • BQ: 4 (1.3%)
  • Green: 1 (0.3%)

The results of the popular vote (after removing independents, for reasons that I’ll explain later) were:

  • Conservative: 39.8%
  • NDP: 30.8%
  • Liberal: 19.0%
  • BQ: 6.1%
  • Green: 3.9%
  • Christian Heritage: 0.13%

(12 other minor parties are omitted, because it turns out they don’t get a seat when all is said and done.)

Assume that the total number of seats to be allocated is 400, of which 308 have already been allocated (as district seats). That leaves 92 seats to be allocated using the Sainte–Laguë method (used in New Zealand, for example – I’ll explain how it works and why I support it in a moment). I’m not going to detail the math yet, I’m just going to show the results:

Party District seats % List seats Total seats % Deviation from proportionality
Conservative 166 53.9% 0 166 40.9% +1.1%
NDP 103 33.4% 20 123 30.3% −0.5%
Liberal 34 11.0% 42 76 18.7% −0.3%
BQ 4 1.3% 22 24 5.9% −0.2%
Green 1 0.3% 15 16 3.9%
Christian Heritage 0 1 1 0.2% +0.1%
Totals 308 100.0% 98 406 100.0%

Note that even with almost a hundred extra seats, it’s still not enough to correct the Conservatives’ undeserved majority (which they got by exploiting the quirks of FPTP). We end up with an overhang of 6 seats.

Nevertheless, look at how much closer to proportional we got! This is what the deviation from proportionality looks like under the current system, and under my proposed system:

Party Real deviation Ideal deviation
Conservative +14.7% +1.1%
NDP +2.8% −0.5%
Liberal −7.9% −0.3%
BQ −5.1% −0.2%
Green −3.6%
Christian Heritage −3.6% +0.1%

Look at that! We went from a deviation of almost fricken’ 15% for the Conservatives… to barely more than a percent. Only the Conservatives have a deviation of more than half a percent from the number of seats they deserve, and only that because we didn’t have enough list seats so we ended up with an overhang. In the case of the Liberals, the Greens and Christian Heritage, the deviation is less than a single seat, as it should be. And it’s just barely more in the case of the NDP, but that’s an artifact of the Conservatives having so many district seats.

If we’d added 50 more list seats to bring the total of seats up to 450, the overhang would be eliminated. Here is what the results would look like then:

Party District seats % List seats Total seats % Deviation from proportionality
Conservative 166 53.9% 13 179 39.8% −0.04%
NDP 103 33.4% 36 139 30.7% +0.11%
Liberal 34 11.0% 52 86 19.1% +0.11%
BQ 4 1.3% 23 27 6.4% −0.07%
Green 1 0.3% 17 18 4.0% +0.07%
Christian Heritage 0 1 1 0.2% +0.09%
Totals 308 100.0% 142 450 100.0%

As expected, no party’s deviation is larger than a single seat. In fact, no party’s deviation is larger than half a single seat. I had to add another decimal place to the deviations just to make them noticeable.

You may be wondering about the fact that the Christian Heritage party got a seat. Where the hell did they come from, right? Well, that’s an artifact of the Sainte–Laguë method I’ve chosen. I’ll explain why in a moment.

So what is the Sainte–Laguë, and why do I like it?

Let’s start from what the situation would look like after voting is done. The district representatives would have been chosen in their individual district elections, and the list votes from country-wide totalled and counted. Using the 2011 results with a hypothetical 400 seat parliament, 308 district reps have been elected (the 308 winners from 2011), and an additional 92 party list seats are to be filled. Because there wasn’t actually a party list vote, I’ll be pretending the popular vote was the party list vote (and ignoring the votes for independents, because they don’t have parties).

District reps (308 seated):

  • Conservative: 166
  • NDP: 103
  • Liberal: 34
  • BQ: 4
  • Green: 1

List vote (92 to be seated):

  • Conservative: 5,832,401
  • NDP: 4,508,474
  • Liberal: 2,783,175
  • BQ: 889,788
  • Green: 576,221
  • Christian Heritage: 19,218

(Again, I’m ignoring the other 12 parties because they don’t get a seat in the end.)

What we need to do now is allocate the remaining 92 list seats, and we need to do it in a way that leaves us with the proportions of the 400 seats as close to the list vote proportions as possible.

There are two broad categories of solutions for doing that: highest averages methods, and largest remainder methods.

The largest remainder methods involve basically determining seat quotas. The simplest and most obvious example is to just to proportion the 400 seats by the list vote – so the Conservatives would get 159 (39.8% of 400), the NDP would get 123 (30.8% of 400), and so on – then top up the seats for those parties lacking using those proportions. But while it sounds obvious and simple, in practice you get all kinds of weirdness using these methods. In fact, it has been mathematically proven that you must get weirdness with quota systems, whenever you have 3 or more parties. For example, the Alabama Paradox was discovered in 1880 by accident, when the chief census clerk noticed that if there were 299 seats in the US House then Alabama would get 8… but if there were 300 seats then Alabama would get 7.

Highest averages methods sacrifice the mathematical purity and logic of quota systems, but in exchange produce results that, in practice, are much more sane and sensible. They are iterative methods, meaning you don’t just do one calculation and boom you have all your seats proportioned. Instead you redo the calculations for each seat and each party, taking into account desired proportions and seats already filled by previous steps. So for our example you’d need to do the calculation 7,200 times (400 seats × 18 parties). Obviously that’s a pain to do by hand, but this is the 21st century. Who is doing electoral calculations by hand anymore?

There are several highest averages methods. The most popular is the D’Hondt method, and of course the method I’m recommending is Sainte–Laguë, and there are others. The difference between the methods is in how the quotient for each party (for each seat) is calculated. Each method uses a different divisor formula.

For example, D’Hondt uses:

\frac{V}{s + 1}

Where V is the number of votes the party received (or the proportion – works either way), and s is the number of seats they have been allotted so far. The divisor values this produces are 1, 2, 3, 4… basically the natural numbers.

Sainte–Laguë uses:

\frac{V}{2 s + 1}

The divisors are 1, 3, 5, 7… basically the odd numbers.

Other variations are {2 s + 2}, which gives {2, 3, 4, 5,…} (it skips 1), \sqrt{n \left ( n + 1 \right )}, etc.. There is a variation of Sainte–Laguë called the “modified Sainte–Laguë method”, used in Norway, that substitutes 1.4 for the first number (so {1.4, 3, 5, 7,…}).

What difference does the divisor make? Well, the larger the divisor, the harder it is to get a seat. Basic math knowledge tells us that the bigger the number on the bottom is, the smaller the quotient (¼ is smaller than ½). In this calculation, the number on top is constant for a party (it is the number/proportion of votes they got) while the number on the bottom grows each time they get another seat. (That’s how it creates proportionality – the more seats you have already earned, the harder it is to earn more, subject to the constant on top, which is the number of votes you got.)

That means that variations that grow the divisor more quickly favour smaller parties. Sainte–Laguë ({1, 3, 5, 7,…}) is more likely to give seats to smaller parties than D’Hondt ({1, 2, 3, 4,…}), because the more seats you have, the harder it gets to get another seat.

It also means variations that make the first number larger make it more difficult to get that first seat. The modified Sainte–Laguë method substitutes 1.4 for the first divisor when s is zero (rather than 1 in the “normal” Sainte–Laguë method) to make it harder for tiny parties to get a single seat. The goal is avoid a smattering of really tiny parties each managing to squeak a single seat (which would ultimately make them rather pointless in Parliament), but if you can manage to earn one, you probably deserve it (and you get a good shot at a second). The Christian Heritage party gets a seat with the Sainte–Laguë, but with the modified Sainte–Laguë method, they don’t – the NDP gets it. (In the 450 seat case, the Conservatives get it.)

The reason my ideal system uses Sainte–Laguë is precisely because it gives smaller parties more of a chance. This is not because I have a hate-on for large parties – quite the opposite, I would love nice big party allocations in Parliament, and failing that, broad coalitions. No, the reason is philosophical, based on the nature of Canada. In smaller, more homogenous countries – like Denmark and Finland and the Netherlands – it makes sense to give more weight to bigger, more popular parties. But Canada is huge, and politically, ethnically, and geographically diverse. Giving preference to large parties is basically handing control of Canada over to what Ontario wants (or, if Ontario can’t find consensus, then Québec), or what white Canadians with European heritage want. Giving more of a shot to smaller parties is healthier for Canada’s diverse concerns, and if they can manage to drum up enough votes to make a dent, proportion-wise, overall, I’d say they deserve a seat.

That being said, I’m not keen on the idea of a dozen small parties getting a single seat each. My idea for more participation of smaller parties is more along the lines of two or three parties with two or three seats each. The Sainte–Laguë method makes it easier for smaller parties, but it also means a good chance of a bunch of tiny parties getting one seat each.

For that reason, there is the modified Sainte–Laguë method, where the first divisor is inflated. With the standard Sainte–Laguë method, the first divisor – when a party has zero seats – is 1. The modified Sainte–Laguë method uses 1.4 instead. This means it’s a little harder to get that first seat. But it also means that if you do get it, you’re much closer to getting a second seat than you are under the unmodified Sainte–Laguë method.

There’s also another issue: thresholds. Many proportional systems use thresholds to exclude tiny parties; if you can’t get, say, 5% of the total vote, you’re simply not included in the proportionality calculations. Again, the idea is to avoid giving single seats to dozens of tiny parties. If the threshold is too high, then it’s really just favouring huge parties… but if it’s low enough, but not too low, it will block out parties that would only win a single seat, but allow parties that will win two or three, which is the scenario I’m looking for.

Deciding whether to use the modified Sainte–Laguë method or thresholds is a tricky and complex decision. We can’t base it on what we observe in past elections, because switching to a PR system will radically change the way Canadians can select their government. Right now most of the major parties are specifically designed to exploit the imbalances inherent in FPTP. If that goes away, they could fall apart entirely, and wholly new parties – and types of platforms – could emerge. I can’t predict on my own how that might unfold. Since I can’t predict what effect they would have, or even if they’re needed, I will abstain for now from taking a position on whether to use the modified Sainte–Laguë method or thresholds.

Thus, my choice is Sainte–Laguë, no modifier, no threshold. Though I am definitely open to convincing on the latter points.

I will do a demonstration later in the post. The system and it’s calculations are not at all hard – it’s all elementary math – but they are tedious. You’ll want to do it with a spreadsheet or program, rather than by hand (and the program/spreadsheet isn’t all that complex).

The ballot

So what would my ideal electoral system look like for the voters? The easiest way to illustrate that is by showing you a hypothetical sample ballot.

The ballot would have two sections – one to select the riding representative, and the other to select the party you want to lead the government. The first section would allow preferential ranking. The second section would allow only a single selection.

I’m just going to do it in English, because I don’t presume any Francophones read this blog, but of course the ballot would have to be available in all major languages: English, French, and in certain areas maybe even Inuktitut. I suppose this could be done by trying to print bilingual or multilingual ballots, but it seems to me that it would be easier to just have the same ballot available in multiple language. If they’re still using paper ballots rather than electronic voting machines, perhaps they could be printed on demand.

I’ll use my riding – Burlington – in 2011 federal election, to demonstrate a sample ballot. It might have looked like this:

[A sample ballot for the hypothetical electoral system I am proposing. It is divided into two sides. One side lists the names and parties of the five 2011 federal candidates in the Burlington electoral district. The instructions tell voters to rank them in order numerically (1, 2, 3,...) in order of preference, noting that they do not need to rank all candidates. The other side lists all eighteen political parties that contested the 2011 federal election. The instructions tell voters to place an X by one of them.]

To me, even though that ballot requires making two votes, it seems pretty straightforward.

Here’s how I might have filled it out in the 2011 election:

[A sample filled ballot for the hypothetical electoral system I am proposing. On the district candidates side, three of the five candidates are ranked as 1, 2, and 3. The remaining two candidates are not ranked. On the party list side, one party is marked with an X. The party that is selected is the party of the candidate that was ranked #2.]

As you can see, I wouldn’t have bothered to rank all 5 candidates. Also, my personal choice for representative for my district would not have been the same as the party I wanted to form government. Brierley was the only candidate who actually came to my door. However, I did not want a Liberal government, I wanted an NDP government. On the other hand, I never saw or heard from Laird once during the campaign. So either I would have had to choose the good representative with a shitty party, or the good party with a shitty representative. But with MMP, that’s not a problem. I can choose the good candidate, and the good party, even if they’re not related.

Here’s another way someone might have filled out the ballot:

[A sample filled ballot for the hypothetical electoral system I am proposing. On the district candidates side, the five candidates are ranked as 1, 2, 3, 4, and 5. On the party list side, one party is marked with an X. The party that is selected is not the party of any of the five district candidates.]

This person ranked all 5 candidates… but then chose a party that didn’t even have a candidate in the riding! That’s pretty cool. It means that even though the only options in my district were the big 4 national parties and 1 person from the Marxist–Leninist party, I could give a vote to, say, the Marijuana Party, or the Pirate Party. That’s really awesome because it means parties whose platform doesn’t really work with district-focused representation – like issues parties – can get votes without having to run candidates across Canada.

You don’t necessarily need to rank the candidates if you only have one choice in mind:

[A sample filled ballot for the hypothetical electoral system I am proposing. On the district candidates side, one of the five candidates is marked with a check mark. On the party list side, one party is marked with a check mark. The party that is marked on the party list side is the same party as the chosen district candidate's party.]

That’s a perfectly legitimate ballot.

Of course, if they had put marks (not numbers) in two candidate boxes, or two party boxes, that would have spoiled it. Likewise if they put the same number in two candidate boxes. If they used non-consecutive numbers while ranking the candidates, or didn’t start from 1, the ordering should be respected as if they were ranked starting from 1. If they tried ranking the parties, I’m not sure whether to count that as a vote for the first-ranked party, or to call it spoiled. I imagine Elections Canada has a policy already in place for current ballots. In any case, if they spoil (or leave empty) only one side of the ballot, the other side should still be respected. Electronic voting systems would make much of this automatic and mindless – they wouldn’t let you choose more than one party, and they wouldn’t let you make mistakes like giving the same rank to two candidates.

So while the ballot is a little more complex than what is currently use in our non-proportional FPTP elections, I think it’s still quite easy to understand. And the extra complexity is well-justified. Voters will get far more flexibility to express their desires than they do now. But if they don’t want it, and just want to pick a local representative and be done with that, they’re free to do that, too.

Example election

Okay, so I’ve talked at length about what I want and why. I’ve even shown you a sample ballot. Now I’d like to demonstrate how the election might actually be carried out, from party registration, to voting, to counting.

I will have to demonstrate the counting in two stages, because the MMP system I propose works in two stages. First I will demonstrate district representative selection by using a single, imaginary riding as an example (I can’t do it with any real data because past elections have not used preferential ballots). Then I will demonstrate how the Sainte–Laguë creates proportionality via extra list seats.

I will not be specifying everything about the process. I will gloss over details that don’t have a major impact on the design. Nevertheless, I hope I’ll give enough detail to make the idea clear.

So, let’s get started.

Party registration

Once the writ is dropped, every political party will have to register for the election. So too will every candidate in every party, as well as all independent candidates. Elections Canada already has procedures for doing this, and there is no reason why they would need to change.

There is, however, one slight difference. When parties register, they must not only provide the list of which candidates they are running in which ridings, as they do now. They must also provide a list of candidates in order of seating preference. The seating preference list decides which candidates will get any list seats the party wins, in order from first to last. So if the party wins 10 list seats, the top 10 names on their seating preference list will be seated, if they are not already.

I don’t see any reason that a candidate cannot both run for a district seat and be on the party’s seating list. If they win the district seat, they will simply be ignored when it comes time to allocate list seats – the next unseated person on the list will get the seat.

So let’s use the 2011 federal election as an example, and the Liberals. For reasons that will become clear shortly, the number of seats up for grabs will be 400 – 308 district seats (the ones that actually existed in the 2011 election) plus 92 list seats (for the purposes of the example).

The Liberal Party of Canada would register themselves as a political party, and the 308 candidates they want to run for the 308 ridings across Canada – that part of the process is unchanged. But they would also supply a list of candidates for list seating preference. They could simply supply the same 308 names that are running for district seats, maybe with 92 extra names to fill it out. Or they could supply 400 completely new names. Or they could mix and match – some of the names may also be running for district seats, some may not.

Presumably what they would do is put the party’s top people at the top of the seating preference list. Right at the top would be Ignatieff. The logic behind that is simple. Unless you really get hammered at the polls, or you do incredibly well in winning district seats but incredibly poorly in winning the popular vote (which happened to the Conservatives in 2011), you are almost certain to win at least a couple list seats. As you’ll see in the final results, the only party with more than half a percent of the popular vote and didn’t get a list seat was the Conservatives… but that was because they won so many more district seats than they by right deserved. Even the Green Party gets 15, and the Liberals got 42. So if you’re up near the top of a major party’s list, you’re pretty much guaranteed a seat.

Ignatieff ran for a district seat – Etobicoke—Lakeshore. If he won that seat, fine, he gets to sit in Parliament. But if he loses the district… which he did… if he’s at the top of the seating list, he’ll almost certainly still get a seat. Which he would have. And that’s a good thing – parties want to get their top people seated, even if they fail to win a district. And they even have the option of not even running in one. (And it might even be a good idea to have the party leadership not run for district seats, so they can focus on bigger issues and not on representing a district.)

So the Liberal Party of Canada would submit their own party registration, 308 registrations for the candidates running in the 308 districts, plus a list of candidates – likely with Ignatieff and other party big shots at the top – in order of seating preference. And there may be overlap between the names on the seating list, and the candidates running in the 308 districts.

I don’t think Elections Canada should require some minimum number of names on the seating list. There are, after all, parties with very, very few candidates. One party in 2011 was only a single dude. I would say that if a party wins more list seats than there are names on their list, then what they have to do is use the lists of other parties. For example, if a party with 5 names on their list wins 8 list seats, then they can tell Elections Canada to give 2 seats to, say, the NDP, and 1 to the Greens. So the next 2 unseated names on the NDP‘s list, and the next one on the Green’s list, get seated. Either that, or the rule should be that the seats will just remain empty. Or maybe a combination of the two rules.

One other thing that might have to change is that it should be possible for a party to have no district candidates, but supply a seating preference list. This would allow parties that don’t actually run on local representation platforms, but rather on issues platforms. Obvious the party would win no seats in the district vote, but if they get enough votes on the party list vote, they might still win seats.

Independents will register as usual. Not having a party, none of the list stuff involves them. They just register for a specific district as an independent.

So, in summary, every party will have to register, and submit two things: the list of candidates running in districts (and which district they’re running in) as they do now, and also a list of candidates to be seated in order in list seats. Every independent will simply register as a candidate in their district.

Voting

On elections day, Canadians go to the pols as always, except the ballots they are handed are somewhat more complex. Rather than just a list of candidates, the ballot now has two sides: a list of candidates who want to represent their district on one side, and a list of all parties running in the election on the other. This is basically the ballot I showed previously. You rank the candidates, and select a party. Simple.

The district vote: Alternative vote in action

Let’s use Elizabeth May’s district – Saanich—Gulf Islands – as our example.

There are 4 candidates:

  • Gary Lunn (Conservative, incumbent)
  • Elizabeth May (Green)
  • Renée Hetherington (Liberal)
  • Edith Loring-Kuhanga (NDP)

There were 91,673, but only 68,987 voted. 160 ballots were rejected, so the total vote count was 68,827.

Now in reality, voting was done with simple ballots – you just marked an “X” next to the candidate you wanted. But under my system, the ballots are preferential. You can still just mark an “X” if you want to… but you also have the option of voting for multiple candidates by ranking them in order of preference.

Let’s play a hypothetical and assume that everyone who voted for Hetherington put Loring-Kuhanga as their second choice… and half of them put Lunn as their third choice. Also, everyone who voted for Loring-Kuhanga put Lunn as their second choice.

So there are 5 types of ballots:

  • 31,890: 1) May.
  • 24,544: 1) Lunn.
  • 8,185: 1) Loring-Kuhanga, 2) Lunn.
  • 2,104: 1) Hetherington, 2) Loring-Kuhanga, 3) Lunn.
  • 2,104: 1) Hetherington, 2) Loring-Kuhanga.

On the first count, the results are everyone’s first choice:

  • 31,890: May.
  • 24,544: Lunn.
  • 8,185: Loring-Kuhanga.
  • 4,208: Hetherington.

Nobody has a majority (50% + 1 at least), so nobody wins yet. The loser, though is Hetherington. She gets eliminated. The ballots that named her first choice are redistributed, looking at second choices. So now the ballots available look like this:

  • 31,890: 1) May.
  • 24,544: 1) Lunn.
  • 8,185: 1) Loring-Kuhanga, 2) Lunn.
  • 2,104: 1) Hetherington, 2) Loring-Kuhanga, 3) Lunn.
  • 2,104: 1) Hetherington, 2) Loring-Kuhanga.

Now the results are:

  • 31,890: May.
  • 24,544: Lunn.
  • 12,393: Loring-Kuhanga.

Still no one has a majority, so the lowest ranked candidate – Loring-Kuhanga – is eliminated, ballots for her redistributed:

  • 31,890: 1) May.
  • 24,544: 1) Lunn.
  • 8,185: 1) Loring-Kuhanga, 2) Lunn.
  • 2,104: 1) Hetherington, 2) Loring-Kuhanga, 3) Lunn.
  • 2,104: 1) Hetherington, 2) Loring-Kuhanga.

Note that the last set of ballots has been completely exhausted, so they are removed:

  • 31,890: 1) May.
  • 24,544: 1) Lunn.
  • 8,185: 1) Loring-Kuhanga, 2) Lunn.
  • 2,104: 1) Hetherington, 2) Loring-Kuhanga, 3) Lunn.

Now the results are:

  • 31,890: May.
  • 34,833: Lunn.

Lunn has a majority, so he wins.

Of course, that’s not what actually happened (nor was it ever likely to happen). In reality, Elizabeth May won. What I wanted to illustrate, though, is that if you have strong support as a second choice, it is possible to come from behind. May led in the first two rounds of the example above, and only lost on the third.

Now, alternate vote likely wouldn’t have made any difference in May’s case. It’s almost certain that more people who voted for Loring-Kuhanga or Hetherington would have had May as a second or third choice before Lunn. (And even if none of them had May as an alternate choice, so long as less than 7,346 out of 12,393 NDP and Liberal voters didn’t have Lunn as an alternative, May would have won anyway.) But there are other ridings where it just might have made a difference.

This same process is repeated for all 308 electoral districts. For the purposes of the example, I’m just going to assume the same 308 winners as those we actually got in 2011.

Getting proportionality via the list vote: Using Sainte–Laguë method

Obviously there was no real party vote in 2011. So I will just… pretend there was, and use the 2011 popular vote numbers as the list vote. I’ll just ignore the votes cast for independents (because they don’t have parties).

So, that means the list vote results are:

Party Votes Proportion
Conservative 5,832,401 39.817%
NDP 4,508,474 30.779%
Liberal 2,783,175 19.001%
Bloc Québécois 889,788 6.075%
Green 576,221 3.934%
Christian Heritage 19,218 0.131%
Marxist–Leninist 10,160 0.069%
Libertarian 6,017 0.041%
Progressive Canadian 5,838 0.040%
Rhinoceros 3,819 0.026%
Pirate 3,198 0.022%
Communist 2,925 0.020%
Canadian Action 2,030 0.014%
Marijuana 1,864 0.013%
Animal Alliance 1,451 0.010%
Western Block 748 0.005%
United 294 0.002%
First Peoples National 228 0.002%

If this were being done properly, you would have to calculate the Sainte–Laguë quotients for every party in the election – at least, every party that passes the threshold, if any. Spoiler alert, the only parties that will win seats are the top 6. So I’ll just ignore the other parties. (Or you can imagine that there’s a 0.1% threshold, which eliminates all other parties from consideration.)

The formula for calculating Sainte–Laguë quotients is:

q = \frac{V}{2 s + 1}

Everyone starts with 0 seats. The vote proportions (using the 2011 popular vote) are:

  • Conservative: 5,832,401
  • NDP: 4,508,474
  • Liberal: 2,783,175
  • BQ: 889,788
  • Green: 576,221
  • Christian Heritage: 19,218

So these are the values for V for each party. They are constant throughout the whole process.

First the Conservatives. Their quotient is:

q_{Con}=\frac{V_{Con}}{2 s_{Con} + 1}
q_{Con}=\frac{5,832,401}{\left ( 2 \times 0 \right ) + 1}
q_{Con}=5,832,401

The NDP quotient is:

q_{NDP}=\frac{V_{NDP}}{2 s_{NDP} + 1}
q_{NDP}=\frac{4,508,474}{\left ( 2 \times 0 \right ) + 1}
q_{NDP}=4,508,474

And so on for the other parties. (Note that in the first round, the quotient is just equal to their vote proportions. That’s because the first divisor is 1. If we used the modified Sainte–Laguë method, that would not be the case.)

So the first round result is:

Round Conservatives NDP Liberals BQ Greens Christian Heritage Winner
Seats Quotient Seats Quotient Seats Quotient Seats Quotient Seats Quotient Seats Quotient
0 0 5,832,401 0 4,508,474 0 2,783,175 0 889,788 0 576,221 0 19,218 Conservatives

The Conservatives win the round because they had the highest quotient, so they get a seat:

Round Conservatives NDP Liberals BQ Greens Christian Heritage Winner
Seats Quotient Seats Quotient Seats Quotient Seats Quotient Seats Quotient Seats Quotient
0 0 5,832,401 0 4,508,474 0 2,783,175 0 889,788 0 576,221 0 19,218 Conservatives
1 1 ??? 0 4,508,474 0 2,783,175 0 889,788 0 576,221 0 19,218 ???

Now the Conservatives’ seat count has changed, so we have to recalculate their quotient:

q_{Con}=\frac{V_{Con}}{2 s_{Con} + 1}
q_{Con}=\frac{5,832,401}{\left ( 2 \times 1 \right ) + 1}
q_{Con}=\frac{5,832,401}{3}
q_{Con}=1,944,134

Which gives:

Round Conservatives NDP Liberals BQ Greens Christian Heritage Winner
Seats Quotient Seats Quotient Seats Quotient Seats Quotient Seats Quotient Seats Quotient
0 0 5,832,401 0 4,508,474 0 2,783,175 0 889,788 0 576,221 0 19,218 Conservatives
1 1 1,944,134 0 4,508,474 0 2,783,175 0 889,788 0 576,221 0 19,218 NDP

In this round, the NDP have the highest quotient, so they get a seat:

Round Conservatives NDP Liberals BQ Greens Christian Heritage Winner
Seats Quotient Seats Quotient Seats Quotient Seats Quotient Seats Quotient Seats Quotient
0 0 5,832,401 0 4,508,474 0 2,783,175 0 889,788 0 576,221 0 19,218 Conservatives
1 1 1,944,134 0 4,508,474 0 2,783,175 0 889,788 0 576,221 0 19,218 NDP
2 1 1,944,134 1 ??? 0 2,783,175 0 889,788 0 576,221 0 19,218 ???

Now we have to recalculate the NDP‘s quotient:

q_{NDP}=\frac{V_{NDP}}{2 s_{NDP} + 1}
q_{NDP}=\frac{4,508,474}{\left ( 2 \times 1 \right ) + 1}
q_{NDP}=\frac{4,508,474}{3}
q_{NDP}=1,502,825

Which gives:

Round Conservatives NDP Liberals BQ Greens Christian Heritage Winner
Seats Quotient Seats Quotient Seats Quotient Seats Quotient Seats Quotient Seats Quotient
0 0 5,832,401 0 4,508,474 0 2,783,175 0 889,788 0 576,221 0 19,218 Conservatives
1 1 1,944,134 0 4,508,474 0 2,783,175 0 889,788 0 576,221 0 19,218 NDP
2 1 1,944,134 1 1,502,825 0 2,783,175 0 889,788 0 576,221 0 19,218 Liberals

And the Liberals get the next seat.

And on and on it goes for all 400 seats. I’ll just skip ahead and show the last rounds:

Round Conservatives NDP Liberals BQ Greens Christian Heritage Winner
Seats Quotient Seats Quotient Seats Quotient Seats Quotient Seats Quotient Seats Quotient
0 0 5,832,401 0 4,508,474 0 2,783,175 0 889,788 0 576,221 0 19,218 Conservatives
1 1 1,944,134 0 4,508,474 0 2,783,175 0 889,788 0 576,221 0 19,218 NDP
2 1 1,944,134 1 1,502,825 0 2,783,175 0 889,788 0 576,221 0 19,218 Liberals
399 159 18,399 123 18,253 76 18,191 24 18,159 16 17,461 1 6,406 Conservatives
400 160 18,283 123 18,253 76 18,191 24 18,159 16 17,461 1 6,406 Conservatives

So the final proportional seat tallies are:

  • Conservative: 160
  • NDP: 123
  • Liberal: 76
  • BQ: 24
  • Green: 16
  • Christian Heritage: 1

From those values, we subtract the seats they won via their district reprsentatives:

  • Conservative: 166
  • NDP: 103
  • Liberal: 34
  • BQ: 4
  • Green: 1
  • Christian Heritage: 0

Which gives:

  • Conservative: −6
  • NDP: 20
  • Liberal: 42
  • BQ: 20
  • Green: 15
  • Christian Heritage: 1

Now, we have a little problem. The Conservatives won too many seats with their district reps. But we’re not going to take away seats from already elected representatives. We’ll just have to live with the overhang of 6 seats – Parliament will have a total of 406 seats for this session. Oh well, we’ll get by.

Thus, we get the final results I mentioned previously:

Party District seats % List seats Total seats % Deviation from proportionality
Conservative 166 53.9% 0 166 40.9% +1.1%
NDP 103 33.4% 20 123 30.3% −0.5%
Liberal 34 11.0% 42 76 18.7% −0.3%
BQ 4 1.3% 22 24 5.9% −0.2%
Green 1 0.3% 15 16 3.9%
Christian Heritage 0 1 1 0.2% +0.1%
Totals 308 100.0% 98 406 100.0%

Summary

So there it is: Indi’s ideal electoral system.

I’ve left many of the low-level details unspecified, but none of them are game-changers. Some study will be required to determine whether and how to tweak – for example, whether to use thresholds to prevent really tiny parties from getting a single seat, and/or whether to use the modified Sainte–Laguë method with the larger first divisor. It may be necessary to implement these tweaks over a number of election cycles, until we can get some kind of stability that works for Canada.

The ultimate goal of my electoral system is to make it possible for voters to clearly express what they want their government to look like, and to eliminate the need to shenanigans like strategic voting. No vote is “wasted” – even if you want to vote for a dark horse candidate, you can do so while still having an impact by listing a major candidate as your second choice or using your party list vote.

The secondary goal is to take away the power of elected politicians to rig things in their favour for the next election. Gerrymandering would be almost pointless, as would trying to run spoilers. The politics of divisiveness just don’t make sense in a proportional world.

However, all the evidence available is that we can expect to see measurable improvement in the way Canada is governed. The benefits of proportional representation are well-documented.

I don’t mean to imply that there won’t be a period of perhaps difficult adjustment. Our most successful parties are currently organized to stake out watered-down, centrist positions, and to refuse any sign of compromise. Under my proposed system, the former strategy will be less effective, the latter will be outright political suicide. Even in the example I ran using the 2011 election above, which was a highly divisive election among the electorate and an overwhelming win for the Conservatives, the Conservatives only managed to get a minority government. At the time, the Liberals swore they wouldn’t form a coalition with the NDP… well, that shit won’t fly under my system. You cooperate, or you vanish into political obscurity. If Canadians wanted to give a party absolute, uncompromising control of the country, they’d fucking well give it to them. In lieu of that, parties had best learn to find balance.

So, what do you think of my proposed system: a mixed-member proportional representation system, with representatives for each district elected via alternative vote, then proportionality with a party list vote achieved using the Sainte–Laguë method. Comments? Suggestions? Let me know.

CC BY-SA 4.0
My ideal for a reformed electoral system by Indi in the Wired is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.

6 Responses to My ideal for a reformed electoral system

  1. One of the main issues that defeated MMP in the Ontario 2007 referendum was closed lists. The Citizens Assembly didn’t have enough time to properly evaluate all the options, and if they’d had more time they would probably have chosen open list in the final recommendation. Brian Tanguay of the Fair Vote Waterloo chapter and Laura Stephenson analyzed the Ontario referendum results, and summarize the reasons for the outcome. http://irpp.org/research-studies/choices-vol15-no10/

    Another common feature of MMP is to combine several current ridings into a multi-member district. Around 9 members per district has been proposed (although I personally think that’s too many). Then the left side of the MMP ballot can be chosen by STV, with an open list on the right side.

    I recommend people read through Wilf Day’s blog http://wilfday.blogspot.ca/ Wilf is a member of the Fair Vote Canada National Council, and has spent much time re-evaluating the election by applying proportional counts to votes.

    –Bob.

    • Regarding open lists:

      I actually like the idea of open lists in principle, but I chose against them for practical reasons. Most places that use open lists seem to be tiny; the only places I could find with a legislature of comparable size were Brazil, Indonesia, and Sweden. The first two are not exactly shining beacons of democracy.

      As for Sweden, it only uses a party list, not MMP, so it makes more sense that they can afford to put more resources into the party list vote – after all, it’s all they’ve got.

      As I explained in the post, I think the dual-vote design of MMP maps quite logically to Canada’s governance. I am loathe to give that up. And if we’re already doing a two-in-one vote, it seems a bit gratuitous to add a *third* layer to each person’s vote for the sake of an open party list. I mean, look how big the sample ballot is already! I thought I was already adding too much complexity when I went with AV for the district rep side, while most MMP implementations use FPTP.

      Yes, it means parties can put utterly unwanted assholes on the list and they get seats, and that sucks… but we already tolerate them parachuting utterly unwanted assholes into districts they know they’ll comfortably win.

      Regarding large multi-member ridings (MMP using STV+Sainte–Laguë):

      I’m wary of making districts too geographically large. Canada is just too huge and too sparsely populated. It’s way too easy for a district to be so large that the concerns across it vary far too wildly for representatives to practically address. You may end up with wacky situations where, for example, a district has voted for all Conservative reps, but the Conservative reps from the north of the district are feuding with the Conservative reps from the south of the district, because the social, economic, and plain old geopolitical concerns of the two areas are so wildly different.

      Given the choice, I would prefer to add *more* districts, and make them smaller, than to go the other way.

      Which segues nicely to….

      Regarding adding more MPs:

      You may have noticed that I spent a fair bit of time pointing out that we don’t necessarily need to increase the number of MPs. That’s because I totally grok the gestalt. Personally, I’m ambivalent about increasing the number of MPs, but I definitely sense what you mention about the idea being unpopular.

      But what you just mentioned about the number of people per MP is really shocking information. I mean, I grew up in Barbados – a quarter million people and 30 MPs, so ~8 kilopeople per MP. Your number made my jaw drop – I had to do the calculation myself. By my estimate, at the time of the 2011 election we had ~109 kilopeople per MP. If Barbados had the same representation, we’d have *TWO* MPs!!! How would we decide which one is the official opposition?!?

      Yeah, we need more MPs. Unpopular though the idea may be, we seriously need more MPs. I’d prefer to push *that* angle, rather than trying to make do with fewer. I’d use the numbers you pointed out – the people per MP, and the time they get per MP – to highlight the issue.

      ——————–

      So I think that if I were to redo my suggestions right now, my immediate reaction is I probably wouldn’t switch the AV side to STV (with larger, multi-member ridings). And I think I probably wouldn’t go with open lists. (Although, I might support a rule where contesting parties must have their list set by a vote *prior* to the actual election – say, at a party convention, or something – or at least make the process for forming their list transparent.) But I probably would suggest adding many more seats, though I’m torn about whether to do that by adding more districts, or more list seats, or both.

      You’ve given me quite a bit to think about though. I’ll have to chew on it for a while. When I read the Stephenson/Tanguay study, I was intrigued by what Citizen’s Assembly advised for Toronto (closed list MMP with FPTP+Hare and a 3% threshold), As opposed to me (closed list MMP with AV+Sainte–Laguë and no threshold). Also the method Shevek proposed on GNU Social looks *very* interesting.

      Oh, and thanks for the tip about Wilf Day’s blog! I’m going to have to read right through that, start to finish.

      • Open lists where the people are elected by the region, just as we do now for the other side of the ballot, is a very good choice in my opinion. Every party you have on your ballots will not be running in every region, so it need not be a whole lot longer. But I really came to tell you that you and this guy should get together and work this out. https://www.sites.google.com/site/ideasfromgord/house-of-commons/canada-s-electoral-system

        • That Gord guy’s site is intriguing. His model is MMP using AV+Hare, but he does the list proportionality step province-by-province (rather than Canada-wide, like I do). He uses closed lists, but he outlines a procedure for selecting them. I don’t quite get the procedure, but it *sounds* like the idea is just that each party somehow holds a vote amongst its members before the election for who gets on the list.

          So it’s *almost* identical to my own system, other than that it does proportionality province-by-province, and that it uses a largest remainder method (Hare) instead of a highest averages method (Sainte–Laguë). I disagree with the latter, but the province-by-province argument is interesting.

  2. Another factor that defeated the MMP referendum in 2007 was the proposed increase of seats in the legislature. Suggesting that Canada gets more politicians is VERY unpopular. Instead of adding seats, when existing electoral districts are combined into multi-member ridings, about 2/3 of the seats can be assigned to that electoral district, with the remainder allocated to party seats. Although members elected to party seats may get their top-up based on a disproportional outcome in another district, they would still serve in the district from which they came. Although this could result in specific regions with disproportional seat allocations, proportionality would be maintained nationally (or provinicially).

    My personal opinion is that Canada is underserved by the existing members of parliament. With (single-member) electoral districts averaging about 100,000 constituents, each member of parliament could spend 8 hours a day, five days a week meeting with constituents, and each constituent would only get about 1 minute a year. That’s not very good representation. And, of course, the reality is that politicians spend far less time than that actually meeting constituents.

    –Bob.

  3. My reason for having 9-10 member multi-member electoral districts is to avoid ridiculously geographically large districts. This mostly works in urban and some rural areas. But Nunavut is a plenty large district all by itself, and it wouldn’t make sense to combine it with other large districts.

    Your comment about pre-selecting the party lists is actually Brian Tanguay’s preferred method too, he called them “Primaries”. But if there’s anything more distasteful to Canadian voters than more politicians, it’s more elections, and what appears to sound like Americanized elections at that.

    I confess, I haven’t looked at the numbers you provide, and now there’s more numbers in that blog 2jenn points us to. And on top of that, I’m supposed to be analyzing and publishing the results of the FairvoteWRC mock elections from the last two festivals (MMP at the Multicultural Festival, and Ranked Ballots at Nonviolence and Kultrun).

    –Bob, who’s been told not to mock elections because they’re election simulations…

Leave a Reply

%d bloggers like this: